This PDF is a selection from an out-of-print volume from the National
Bureau of Economic Research
Volume Title: The Analysis of Public Output
Volume Author/Editor: Julius Margolis, ed.
Volume Publisher: NBER
Volume ISBN: 0-87014-220-8
Volume URL: http://www.nber.org/books/marg70-1
Publication Date: 1970
Chapter Title: Administrative Decision Making and Pricing: Externality
and Compensation Analysis Applied
Chapter Author: Oliver Williamson
Chapter URL: http://www.nber.org/chapters/c3352
Chapter pages in book: (p. 115 - 138)
ADMINISTRATIVE DECISION MAKING
AND PRICING: EXTERNALITY AND
COMPENSATION ANALYSIS APPLIED
OLIVER E. WILLIAMSON
University of Pennsylvania
This paper deals with an administrative decision-making and pricing
problem which, by itself, is so special as almost to be unique. It involves the prospective interference in the reception of television signals
that will result from the construction of the World Trade Center in
New York City. Despite its uniqueness, the issues which it poses are
common to a much wider class of administrative decision-making
and pricing problems. Certain of these are issues which, in the past,
have been incompletely analyzed.
The theoretical area into which this prospective television interference problem falls is that of externality analysis. Three general
issues are posed: First, Itw does the stipulation that spillover costs
be explicitly recognized inftuence project design? Although the answer
to this is straightforward, that to the next question is somewhat more
subtle. Given any project design, what are the allocative efficiency
implications of paying compensation for damages? Determination of
the optimal policy here requires an extension of the existing theory of
externalities to allow for a comparison between what Frank Michelman
has referred to as "demoralization costs" on the one hand and the
administrative costs of paying compensation on the other. Once the
relevant model has been devised, the final question becomes: Under
what circumstances are demoralization costs apt to be especially great?
A fourth issue, but one which is discussed only very briefly, concerns
the "necessity" for centralized review of administrative decision
making so as to assure sensitivity to spillover conditions.
NOTE. Research on this paper was supported by a grant from the National
Science Foundation. I would like to express my appreciation for comments re-
ceived from R. M. Cyert and others when I presented an earlier version of
this paper at seminars at M.I.T. and the University of Wisconsin.
1
F. I. Mich&man, "Property, Utility, and Fairness: Comments on the Ethical
Foundations of 'Just Compensation' Law," Harvard Law Review, April 1967,
Vol. 80, pp. 1165—1258, especially pp. 1208—1218.
j
r
The Analysis of Public Output
116
The paper is divided into four parts. First, the nature of the problem and the prevailing institutional realities are briefly summarized.
The analytical framework within which to examine spillover conditions of the sort described is then developed in the second section.
This framework is applied to the television interference problem in
Section III. The conclusions follow in the last section.
I. The World Trade Center Problem and an
Administrative Solution
On January 8, 1968, Mayor Lindsay's Advisory Task Force on CATV
and Telecommunications (henceforth referred to as the Task Force)2
submitted an interim report to the Mayor dealing with the problem of
television interference posed by the construction of the World Trade
Center.3 The essence of this report, as it bears on the spillover and
compensation question, is as follows:4
The World Trade Center, now under construction by the Port of
New York Authority, will include twin towers 1,350 feet high. The
towers will be erected separately, at an interval of eight months. During part of their construction, the towers will reflect the television signals now transmitted from the Empire State Building. The reflection
will
cause objectionable interference with television reception
in parts of Manhattan and the Bronx. There are conflicting estimates
on the number of television homes that will be affected—ranging
between 100,000 to 600,000 homes. The interference will commence
early in 1970 and will last for at least a year, and perhaps for as much
.
.
.
as two years.
When the north tower of the Center is near completion, the broadcasters will make it their regular television transmitting site, with new
facilities embodying the latest developments in the state of the art.
Once actual transmission from the north tower begins, the objection-
able interference is expected to disappear, and the Task Force assumes
that thereafter there will be improved television reception in the City.
The Task Force has examined a variety of proposals for the solution
2 The author served as an economic consultant to the Task Force. Except for
references to the Task Force report, the views expressed here are my own and
do not necessarily reflect those of the Task Force.
3 A summary appears in the New York Times of January 13, 1968.
The Mayor's Advisory Task Force on CATV and Telecommunications, A
Report on the World Trade Center and Television Reception in New York City,
January 8, 1968, pp. i—u.
A
T
Administrative Decision Making and Pricing
117
of the problem of objectionable interference. One proposal was that
the height of the towers be restricted, but because the City Government can legally neither impose nor enforce such a restriction, it was
not considered a solution. In the end, the Task Force found that the
only possible solutions are (1) the use of directional transmitters
and/or translators and (2) cable television. Neither offers a complete
solution.
In particular, as the report goes on to point out, the capability to
serve all of the affected areas by cable television does not presently
exist and may not by the time the interference develops. In addition,
subscription to cable television at prevailing New York City rates
costs $60 per year. This is not a negligible expense in any case—
especially for the low-income families in Harlem and the Bronx, who
are among those expected to be affected. Both by reason of nonavailability and cost, therefore, cable was not felt to be an adequate
solution to the
The course of action favored by the Task Force, consistent with
the legal restrictions facing the city, was that UHF directional trans-
mitters and/or translators be used to bring a "focused" television
signal into the affected areas. "The frequencies at which
[these]
would operate would have to be in the UHF band because there are
not enough unused VHF channels [the broadcasting stations already
on VHF will continue to transmit at these frequencies, since most of
the City and surrounding areas will not experience the objectionable
interferences in question] and because directional transmitting equipment for the VHF band is too heavy and too large to mount on the
Empire State Building." What is involved, therefore, insofar as the
affected areas are concerned, is a shift of frequencies into unused portions of the UHF band, with directional transmitting used to focus the
signal and thereby avoid interference.
Unfortunately, however, there are real costs to the public entailed
.
.
.
by this solution. For one thing, sets must be able to receive UHF.
It is estimated that at the time the objectionable interference first
begins 25 per cent of all the television sets in the city will lack an
all-channel capability.1 In addition, the Task Force was advised that
an outdoor UHF antenna will be required to receive the directional
signal, and "relatively few
. now have one or are expected to
have one by 1970."s The estimated cost of the antenna is $10, and
for those unable to install their own an additional charge of $12.50
.
ibid., p.
14.
.
8 Loc. cit.
I
118
The Analysis of Public Output
will be involved.9 Thus, even assuming that set replacement is imminent for those lacking an all-channel capability, aggregate costs of
from $2 million to $13 million are involved (depending on which
estimate of the interference cone is used) if families living in the affected areas are to receive an acceptable directionally transmitted
signal.'°
A constitutional bar prohibits compensation in this
which
may go far to explain the neglect of these spillover costs in the initial
design. As the Task Force Chairman, Fred W. Friendly, observed,
the responsibility for considering these spilovers appears to have
"fallen between stools." Lest this situation recur in subsequent administrative decision-making situations, he offered the following suggestions in his transmittal letter to the Mayor:12
At the risk of reciting the obvious, I offer a personal observation
as one new to the problems of City planning. It seems to me that
those who plan a project of large scope, such as the World Trade
Center, must always ask themselves how the project will affect the
total City—how it will infringe upon the urban environment in which
so many millions of us live and work. They must probe the implications of the project for such things as the shape of contiguous parts
of the City, traffic and the movement of persons, aviation safety, and
the construction of similar projects in other parts of the City—as well
as for television reception throughout the City.
Moreover, in a City such as ours, where the activities of its
inhabitants are so complex and interdependent, there must be within
p. 18.
10 This is not
an exhaustive treatment of the spillovers involved, but is
sufficient at least to establish that these are nontrivial in magnitude. Omitted
from the estimate are: (1) prorated set replacements costs; (2) interference that
extends beyond New York City to affect reception in Connecticut; (3) the opportunity costs of time expended to arrange for the installation of an outdoor
antenna (the implicit installation cost assigned to those who install their own is
the commercial charge of $12.50, which is obviously an upper bound); (4) the
possibility that the outdoor antenna will have usefulness that extends beyond the
period of interference. Also neglected is the possibility that, once the antennas
are transferred from the Empire State Building to the World Trade Center, improved reception in much of the city is expected. But perhaps 400,000 homes
will find it necessary to reorient outdoor receiving antennas (at a cost of $7.50
each) at that time also (ibid., p. 11), Ideally, all of these factors are taken account of in the building design; but for our purposes here only the transitional
signal interference problem will be considered.
11 This was the interpretation of counsel to the Task Force of terms in the
New York State Constitution.
12 Letter from Fred W. Friendly to the Honorable John V. Lindsay, January
8, 1968, p. 3.
r
Administrative Decision Making and Pricing
119
our system of public administration a single focus empowered to look
at and to weigh, one against the other, all the implications of a large
project. It may well be that the sum total of the adverse implications
will require that a project be shaped differently, or perhaps even
abandoned, whereas no single negative, looked at in isolation, would
have this weight. In the case of the World Trade Center, authority
has fallen between so many stools—the City, the States of New
York and New Jersey, and two separate agencies of the Federal
Government—that no single body has been able to assume aggregate
responsibility for this task.
II. Efficiency and Demoralization Costs
An efficiency framework for deciding whether to compensate spillovers is developed in Part 1 of this section. Distributional considerations are introduced in Part 2.
1. Demoralization Costs
As indicated previously, the concept of demoralization costs introduced here is attributable to Frank Michelman. It refers to secondary
or adaptive responses taken by those who are made subject to what
they regard as capricious redistributions. Secondary adjustments of
two types are distinguished: protective and aggressive responses. Protective responses involve asset reallocations by members of society
(not necessarily or exclusively the victims) who, observing the cir-
cumstances of capricious redistribution, are anxious to forestall a
similar fate. These reallocations are induced solely by the failure of
compensation to be paid on spillover costs experienced in a nonmarket transaction, and force owners of human or nonhuman assets
to accept lower returns than they could otherwise obtain. Aggressive
adaptations take the form of disruptive acts against society. In response
to what is regarded as a willful destruction of asset values, the victims
and their sympathizers respond in kind.13 Although adaptive responses
'3 A third type of adaptive response not mentioned by Michelman might be
characterized as "despondency." This is demoralization of an extreme sort:
productive activity on the part of losers and their sympathizers is substantially
reduced, and the responsibility for their care may be turned over to the state.
For our purposes here, only adaptive responses of the protective or aggressive
varieties will be treated in the text.
-J
I
120
The Analysis of Public Output
of both types have been noted
they appear not to have
received the systematic treatment that Michelman supplies.
Our purpose in the remainder of this part will be (1) to generalize
the Michelman model and develop the allocative efficiency criteria for
determining when compensation should be paid; and (2) to make
evident the reasons why, even when a project is "optimally" designed,
failure to pay compensation can give rise to an allocatively inefficient
result.'5 The analysis proceeds on the assumption that allocative
efficiency and income distribution objectives can be meaningfully
separated, an aspect of the argument that is examined more thoroughly
in Part 2 below.
Let the scale of a project be given by X; let G(X) be the bene-
fits accruing from the project less the direct costs, S(X) be the assessed
value of the spillover costs if compensation for the full amount of the
losses is paid, D(X) be the demoralization costs (which, as indicated
above, are the secondary responses induced in "uncompensated losers,
their sympathizers, and other observers disturbed by the thought that
they themselves may be subjected to similar treatment on some other
occasion")'6 that are incurred if compensation is not paid, and A (X)
"With regard to protective adaptive responses to uncompensated spillovers,
see 0. E. Williamson, D. G. Olson, and August Ralston, "Externalities, Insurance, and Disability Analysis," Economica, August 1967, Vol. 34, pp. 240—41,
243. In some respects, the model appearing in this paper is merely a simple extension of that developed in the paper just cited. But the present version is more
complete in its statement of motivational assumptions, has greater generality, and
develops the rule-making implications of the analysis in a way that was not ap-
parent to us at the time the Economica paper was written. The possibility of
aggressive adaptive response is referred to in a footnote by Jerome Rothenberg,
The Measurement of Social Welfare, Englewood Cliffs, N.J., 1963, pp. 74—75,
fn. 25.
I neglect the potential misallocative effects which obtain in small numbers
situations when spillover costs are assessed on the agent responsible for the ex-
ternality but compensation is not paid. An opportunity to arrange a bargain
between the parties which shifts the solution away from the social optimum
(judged in allocative efficiency terms) exists in principle in these circumstances.
(See J. M. Buchanan and W. C. Stubblebine, "Externality," Economica, November 1962, Vol. 29, pp. 381—83.) I would not judge this to be a significant factor
in practice.
Michelman, op. cit., p. 1214. Note that what I treat as demoralization costs
is different from Michelman. He also includes what might be characterized as
the immediate experience of disutility borne by losers from the realization that
no compensation will be paid. This is the spillover cost term in my model.
Absent secondary effects (or effects of the sort mentioned in footnote 15, supra),
there would be no necessity for compensation in order to reach an allocatively
efficient result—although, of course, design size should reflect all costs, including
spillovers.
r
Administrative Decision Making and Pricing
121
FIGURE 1
$
+ Ax
x3x2
x0
x
be the administrative costs of determining who is adversely affected
and to what extent. Assume that if compensation is paid at all that it
is paid in the exact amount of the loss experienced. Then the welfare
function can be expressed as:
(1)
V
G(X) — S(X)
— mm
(D(X), A(X)),
and the objective is to choose the optimum project scale so as to
maximize V.'1 This can be shown graphically in Figure 1. Assuming
11 For simplicity of exposition, and so as to focus better on the demoralization
cost issue of principal concern to us here, the argument has been simplified by
omitting two essential aspects of the problem. Thus, whereas we optimize only
with respect to the scale of operations and the decision as to whether or not to
compensate, we should also be sensitive to (1) the mode as well as the scale of
operations and (2) the possibility that spillover can be efficiently mitigated by
taking adaptive action. Implicit in the above formulation is an assumption that
-j
122
The Analysis of Public Output
that all costs manifest themselves as marginal rather than lump sum
values, and treating S1 + D1 and Sx + A1 as vertical summations of
the corresponding marginal terms, this can be interpreted as follows:
The optimum facility size in the absence of external effects is X0.
Given the spillover costs shown by S1. however, externalities begin
to take effect at output X4, the marginal value of which costs increase
monotonicaily with project size. In the absence of demoralization
costs, or if the administrative costs of paying compensation were
negligible (and compensation were actually paid), the optimum facility
size would be X,. Given positive demoralization costs and administrative expense, however, the size must be reduced to X2 or X3 for the
facility to be optimally designed. Whether X2 is to be preferred over
X3 depends on the comparison of over-all net gain, which can be
judged by the reference to the areas designated A1 and A,.'8 If A1
the system has been optimized in both of these latter two respects. Although the
decision of whether or not to compensate sets up incentives such that adjustment
with respect to one of these dimensions will be induced, intervention is necessary
to secure simultaneous satisfaction with respect to both (assuming that bilateral
bargaining is not feasible).
Thus, if compensation is not to be paid, those who sustain the spillover will
have the incentive to adapt against it. Expressing adaptation expense as Z,
their objective, for any given value of X, is to choose Z so as to minimize
S'(X, Z) + Z, where aS'/aZ < 0. If compensation is to be paid, and if spillover
varies with the mode as well as the scale of operations, those who are responsible
for the project will naturally be sensitive to both mode and scale in its design.
In the absence of instructions to the contrary, however (or unless social welfare
considerations reliably prevail), this incentive to select the optimal mode vanishes
if compensation is not required. Those who are responsible for the spillover may
be expected to choose instead the mode that maximized direct net benefits alone:
"excessive" spillover costs are thus apt to be generated. Likewise, the decision to
pay compensation attenuates the incentive of those who experience the spillover
to adapt efficiently against it. A potential breakdown of incentives thus exists
whichever way the compensation decision is made. It should therefore be
stipulated that (1) if compensation is to be paid, damages will be assessed in the
amount necessary to restore wealth to a status quo ex ante condition "as if'
adaptation against spillover has been made, and (2) if compensation is not to be
paid, those who are imposing the externality will be required to select the optimal
mode anyway.
For a discussion of these issues in the context of compensation for accidents,
see Williamson, Olson, and Ralston, op. cit., pp. 237—39. Also see Ronald
Coase, "The Problem of Social Cost," Journal of Law and Economics, October
1960, Vol. 3, pp. 41—42.
18 The total net benefit of operating at X, and paying compensation is given
by the area between G1 and
+ A1 from the origin to X2. The total net
and not paying compensation is given by the area between G1 and S1 + D1 from the origin to X3.
To see that A1 and A, are the relevant areas, let A2 be that part of the shaded
benefit of operating at
area A, to the left of X3 and A," be that part of A, to the right of X3. Now if
c,-.
Administrative Decision Making and Pricing
123
exceeds A2, compensation should not (on allocative-efficiency grounds)
be paid and project size should be limited to X3. If instead A1 is less
than A2, compensation should be paid and the facility should be
extended to size X2. (Obviously if the curve
+
and the curve
are nonintersecting over the relevant region, one always selects
S1 +
the lower of the curves as the appropriate one for determining optimum
facility size, and compensates or not accordingly.)
Note that if the facility is mistakenly designed to size X1 (or larger),
the area A3 (or its counterpart if X exceeds X1) also needs to be
considered in making the decision on whether or not to compensate.
The criterion here is compensate if A2 + A3 is greater than A1, otherwise not. More generally, the following proposition is advanced:
whatever the facility size, if the redistributive consequences of an
activity can be expected to induce secondary responses, failure to
allow for such adaptations can lead to an allocatively inefficient compensation policy. Contrast this with the more common proposition
that once facility size is fixed and its immediate spillover consequences
established, the question of whether or not to compensate involves
only equity but not allocative considerations. But clearly more than a
simple transfer payment is involved if nontrivial administrative costs
will be incurred in paying compensation, and/or if protective or
aggressive adaptive responses to noncompensation are reasonably
probable.1°
It might be protested that this is all well and good, but that it is
equity, not efficiency, that really matters. The vital issue is "what
the facility were built to size X3, the region A1 would represent costs which
could be avoided by not paying compensation, while the area A2' would be
avoidable costs if compensation were paid. All other costs are the same, whatever the compensation decision. Since the benefits, given by the area under G1
from the origin to
are identical in both cases, the choice rests on the relation
of A1 to A21: if A2' is greater than A1 compensate, otherwise not.
The analysis is incomplete, however, because if the decision is made to compensate, the optimal size is X2 not X3. If the size of the facility is extended from
to X2 under the stipulation that compensation will be paid, the additional net
benefit A2" will be realized. Thus the complete criterion is: if A2( A2' + A2")
is greater than A1 operate at X2 and compensate; otherwise operate at X3 and
do not compensate.
19 Uncompensated pecuniary spillovers might also give rise to secondary
adaptive responses which have allocative effects. Relevant in this connection is
Burton Weisbrod's comment on Anthony Downs' paper. Weisbrod points out
that many of the spillovers identified by Downs are pecuniary but not real.
This may be correct in terms of immediate consequences. But despite his
principal emphasis on fairness rather than efficiency, Downs is expressly concerned that noncompensation may induce aggressive adaptive responses—which
is an efficiency aspect of the problem.
I
The Analysis of Public Output
124
ought to be done?"; the above is merely a digression. But this misses
the point. What ought to be done cannot be judged without considering the cost implications. Otherwise, presumably, we would always
compensate—a rule which, manifestly, society has rejected. An
analytical framework is evidently needed to explain this condition.
The above analysis organizes the issues in a way which, if it does not
fully dispose of the compensation question, at least reveals the tradeoffs.
The model can be extended to allow for the possibility of "imperfect" compensation. Thus, let Y be the degree of precision in compensation attempted (where, say, Y is the coefficient of variation in
compensation paid, and a value of Y 0 reflects exact compensation
in the precise amount of the damage inflicted) •20 Obviously, the administrative cost of ascertaining compensation increases as Y decreases.
It also seems plausible to assume that demoralization costs are an
increasing function of Y—either because of risk aversion or because,
when precise compensation is paid, it is evident that society is
earnestly concerned over redressing damages imposed. Both protective
secondary investments and aggressive secondary responses are therefore apt to be an increasing function of Y. Thus the model becomes
(2)
V = G(X)
—
S(X)
—
[D(X, Y) + A(X, Y)J,
and the purpose now is to select X and Y optimally so as to obtain
an allocatively efficient solution. The maximization is again straight-
forward, the difference being that X and Y are now determined
simultaneously, whereas previously Y was arbitrarily set equal to
zero. In the absence of relatively large fixed administrative expenses
(expressed in relation to the extent of demoralization anticipated if
no compensation is paid), compensation (however imperfect) would
become usual on every occasion of spillover. Also note that only if D
and A are additively separable in X and Y will the optimum design
size always be independent of the degree of precision in compensation
attempted.
Consider now the circumstances under which the demoralization
costs resulting from the noncompensation of losses occasioned by
deliberate social action are apt to be especially great. Assuming that
individuals or groups are wealth maximizers, the general proposition
20 Note that a value of Y greater than zero does not imply an undercompensa.
tion bias. It merely reflects a willingness to accept greater variance in the
estimate of losses.
I
Administrative Decision Making and Pricing
125
is this: individuals or groups who (a) either bear or observe the conditions of spillover and (b) can reasonably expect that they will be
similarly disadvantaged in the future have an incentive to adapt in
ways that are calculated to discourage the repetition or attenuate the
effects of similar administrative decisions made subsequently. Within
this framework, the following criteria are advanced as a basis by which
to judge when to compensate. Ceteris paribus, demoralization costs
increase, and thus the case for compensation becomes stronger, as any
of the following happen.2'
1. THE PURPOSIVENESS OF THE ACT AND THE FORESEEABILITY OF ITS CONSEQUENCES INCREASES. The criterion is based
on the assumption that disadvantaged individuals are apt to
feel
especially aggrieved, and hence most apt to adapt against the possibility
of being made to bear such spillover costs at some future date, if it
can reasonably be inferred that they are being deliberately exploited.
The calculated choice to impose spillover costs thus has special counterproductive potency. Unanticipated spillovers, by contrast, are
regarded as essentially random occurrences. The individuals experiencing such spillovers have less cause for believing that they are being
discriminated against, and hence have less incentive to forestall subsequent spillovers by adapting against such contingencies.
2. THE MORE EVIDENT IT IS THAT THE ADMINISTRATIVE
COSTS OF SECURING SETTLEMENT ARE LOW. The assumption
here is that disadvantaged individuals have an appreciation for the
real costs that would be imposed on the state by the stipulation that
all spillovers, however trifling, be compensated. When it is evident,
however, that the administrative costs of securing settlement are low,
failure to pay compensation encourages those who bear the spillover
21 The first five criteria are substantially those proposed by Michelman (op.
cit., pp. 1217—18, 1223). The rationale supplied in each case is mine.
Note that Peter Steiner, in his comment on this paper, contends that "These
criteria are not derived from the theory of demoralization costs." This is true.
But demoralization
cost theory is merely intended as a means by which to
organize the issues and relate this important concept to conventional externality
theory. The criteria for judging when demoralization costs are apt to be es-
pecially great are all derivable from the simple (wealth maximizing) proposition
that people are more prone to adapt when the subjective probability of otherwise
being exploited is perceived to be high. This is, I assume, at least reasonable
if not obvious.
I
•1
126
The Analysis of Public Output
to believe that their property rights have little standing. Hence they
are inclined to adapt (protectively or aggressively) accordingly.22
3. THE GREATER THE HARM EXPERIENCED. When, either
individually or collectively, the harm experienced is large, individuals
and groups who bear the spillover are apt to regard the case for compensation to be especially great. The assumption of decreasing marginal utility of money would tend to support claims of compensation
whenever any individual experiences substantial harm. Spillovers that
are individually small but collectively great, especially if they fall disproportionately on a well-defined minority group in the population,
tend to encourage the view that the group's interests are being subordinated under the prevailing administrative decision-making process.
Hence the inclination to adapt.
4. THE LOWER THE "SIMPLE" NET GAINS [G(x)] THAT THE
PROJECT PRODUCES. The approval of projects for which the
"simple" net gain is small suggests that either special interests or
extra-economic considerations have been accorded great weight. Had
the relevant spillover costs been included, the project might well have
failed to pass the usual economic tests. Again, the apparent subordination of the interests of disadvantaged parties is what stimulates the
adaptive responses.23
5. THE LESS EXTENSIVE IS THE PARTICIPATION IN THE
BENEFITS OF THE PROJECT BY THOSE WHO EXPERIENCE THE
EXTERNAL IT Y. Circumstances for which evident reciprocities between
burden and benefit are lacking are ones which, in the absence of compensation, suggest exploitation of the disadvantaged parties. Adaptive
22 Even if adaptive responses by individuals were independent of the ad-
ministrative costs of settling claims, the case for compensation becomes stronger
as administrative costs decline. This follows directly from an examination of
equation I and Figure 1 above.
23 Kenneth Arrow raised a question at the conference over this criterion. He
suggested that disadvantaged parties might feel especially aggrieved if, for a
project that had enormous net gains (and hence, presumably, the capacity easily
to pay existed), compensation was refused. The point has merit and deserves to
be investigated. Possibly the relationship between compensation and "simple" net
gains is U-shaped.
An additional justification for requiring compensation for marginal projects is
that such a rule would serve to discourage the uneconomic expenditure of
society's resources. Whatever the merits of this judgment, it does not bear
directly on the demoralization question per se.
Administrative Decision Making and Pricing
127
responses calculated to avoid such results in the future are thus encouraged.
6. THE GREATER THE EXTENT TO WHICH THE DISADVANTAGED GROUP HAS PRE-EXISTING GRIEVANCES (BY REASON
OF HAVING BEEN MADE TO BEAR SPILLOVER COSTS IN THE
PAST) AGAINST THE STATE.24 This criterion moves beyond the
consideration of the project taken by itself to consider expected net
consequences over a series of programs. A group that is disadvantaged
in one case, favored in another, unaffected in a third instance, etc.,
may regard the administrative decision-making calculus as one which,
in a composite sense, is "fair." Groups, however, who find themselves systematically disadvantaged have a strong incentive to reduce
their exposure to exploitation.
Since it is costly to apply the above analysis in individual cases to
determine whether, on every occasion of spillover, demoralization costs
are greater than or less than settlement costs, it may be efficient to
develop some crude organizing principles ("rules of thumb"). One
that is commonly recognized by the law is that whenever there is
physical invasion, however small the cost, compensation shall be
paid.25 The reasoning here, presumably, is that the physical taking of
property represents a general threat to secure expectations. Hence,
however trifling the infringement, compensation is paid. A second
rule to deal with situations (such as the case examined above) where
physical invasion is absent might involve a threshold specification.
Whenever the expected aggregate loss or whenever any individual loss
exceeds some specified set of values, the spillover will have its demoralization and administrative costs evaluated. A third would be to
give special consideration to compensation in cases involving previously disadvantaged and aggrieved parties. (Note, however, that it is
probably easier for administrators than for the courts to put such a
rule into effect. For continuing programs involving high exposure to
24 With respect to each criterion, but perhaps especially criterion 6, one would
expect, ceteris paribus. that the probability of aggressive reaction would vary
directly with the degree to which the affected group perceives that the community
is sympathetic to its grievances; grievances that have "legitimacy" in this sense
will, if acted upon, be less vigorously repressed by the exercise of police powers
by the state. Such grievances, eventually, may lead to a legislative correction if
the violation of rights persists.
25 As Michelman points Out, the courts "never deny compensation for a
physical takeover.
This may be true although the invasion is practically
trifling from the owner's point of view: a marginally encroaching sidewalk, for
example" (op. cit., pp. 1184—85).
.
.
.
j
I
128
The Analysis of Public Output
loss of previously disadvantaged parties, legislative action to provide
explicitly for compensation may be warranted.26) Doubtlessly other
"efficient" rules would emerge as experience in implementing the
model proposed is accumulated.
2. Distributional Considerations
Objections to analyses of the above type can and have been made on
grounds that it is inappropriate to separate allocative efficiency from
income distribution.27 This is fundamentally correct. In every allocative
efficiency judgment for which a redistribution is involved there exists
an implicit distributional weighting: usually, that the benefits and costs
are weighted equally "to whomsoever they may accrue." This might
seem to be an insufficiently refined assumption, especially if one is
inclined to the view that "The determination of prevailing values for
a given community . is a proper and necessary task for the
economist."76 Unfortunately, however, the procedures by which the
economist is expected to respond to this injunction and operate upon
.
.
it are not provided.29 It is, perhaps, instructive to note that, immediately following the statement quoted above, Bergson goes on to state
that the necessary value-determination process "is a project which I
shall not undertake here."3°
Lacking a specification of community values, can a case for the
above analysis, with its emphasis on allocative efficiency, nevertheless
be made? At least three arguments suggest themselves. First, for
26 Michelman also recognizes the difficulties that the courts experience with
such cases, and suggests that legislative remedies may be appropriate where
the conditions of such spillover occur repeatedly (op. cit., pp. 1254—56).
27 See, for example, Rothenberg, op. cit., pp. 100—103, and references cited
therein.
28 Abram Bergson, "A Reformulation of Certain Aspects of Welfare Eco-
nomics," Quarterly Journal of Economics, February 1938, Vol. 52, p. 323.
29 An approach that has recently been proposed that has promise
is the
treatment of distributional questions at a rule-making (constitutional) level. On
this see J. M. Buchanan and Gordon Tullock, The Calculus of Consent, Ann
Arbor, Michigan, 1963, Ch. 6; Harvey Leibenstein, "Long-run Welfare Criteria,"
in Julius Margolis (ed.), The Public Economy of Urban Communities, Baltimore,
Maryland, 1965, pp. 39—51; Michelman, op. cit., 1218—24. Conceivably the law
can evolve appropriate rules by framing the distributional issue in a repeated
play context of this sort, but for this one would expect that the allocative efficiency framework developed above can be used as input to such a process. Indeed, as Michelman points out, "we shall find ourselves asking much the same
questions to determine whether a compensability decision is fair as were suggested by the [allocative efficiency] approach" (op. cit., p. 1223).
Bergson, op. cit., p. 323.
c
Administrative Decision Making and Pricing
129
purposes of illustrating the general properties of the model, any distributional assumption will do. (For example, the functions and curves
could be interpreted as ones which already embed the appropriate
distributional weights.) The same types of implications with respect to
a treatment of externalities will obtain.31 Second, if one assumes that
the political process has already "solved" the distributional problem,
one might take the position that, subject to the condition that the
movements involved are small, neutrality at the margin is appropriate.
Third, whether distribution is "correct" or not, it may be impolitic
to employ any valuation scheme other than one which, in the first
instance at least, weights benefits and losses equally. It should be
noted, however, that to proceed in this way does not, if demoralization
costs are introduced in the way suggested above, suppress distributional considerations entirely. Thus whether a program or set of rules
is apt to have serious secondary consequences is a function of who the
affected parties are. Hence, even within what is conceived of as an
allocative
efficiency
framework,
distributional considerations can
manifest themselves in at least a limited way.
If, as seems plausible, the very rich are the ones most apt to make
asset reallocations in response to uncompensated spillovers, while the
poor will be most ready to engage in aggressive behavior, the broad
middle class may be the group least compensated under these rules.
This may appear inequitable, and a remedy therefore indicated. Such,
however, moves outside the allocative efficiency framework herein
developed. Constitutional questions of fairness are involved.
III. Application to the World Trade Center
My first purpose will be to indicate what appear to be the properties
of the World Trade Center design under the prevailing institutional
arrangements. Given this design (and assuming away for the moment
the constitutional bar to compensation), the proposed criteria by
which to judge when demoralization costs are apt to be especially
great are successively applied to the television interference problem.
Finally, the centralized versus decentralized decision-making problem
is briefly considered.
31 An exception might be noted. If, under the prevailing social value system,
no uncompensated spillovers are admissible, the demoralization cost expression
vanishes. Here, optimality always occurs at X2 (where S1 +
intersects
and compensation is always paid.
1
The Analysis of Public Output
130
1. The World Trade Center Design
Inasmuch as the television interference effects (and possible other
height-related spillovers—e.g., air traffic interference) apparently were
not considered when the Center was designed, the value of X0 on
Figure 1 presumably reflects the design height selected of 1,350 feet.
The value of X4 represents the height at which interference first becomes significant, namely 900 feet, and X2 or X3 is the optimal facility
size.32 Which of these two values is allocatively the most efficient
depends on a comparison of administrative and demoralization costs.
Are there genuine secondary effects to be concerned over, and are the
administrative costs of paying compensation too great as to justify
bearing them?33
2. The Criteria Applied
Consider criterion 1: the purposiveness of the act and the foreseeability of its consequences. There is no question that the World Trade
32 "There is general agreement that if the towers were to rise no higher [than
900 feet] there would be no objectionable interference with television reception
in New York City" (Task Force, op. cit., p. 12). There is at least a possibility
that were the Port Authority even now presented with the requirement that
compensation be paid for interference created, redesign might still occur—although this is perhaps doubtful. Inasmuch as the estimated cost of the World
Wide Trade Center is $575 million (ibid., p. 4), redesign might be prohibitively
expensive.
Peter Steiner, in his comment on this paper, poses the question: Is the
concept of demoralization costs operational? My numerous references to Michelman's fundamental examination of this question (op. cit.) did not, apparently,
satisfy him. The answer that emerges from my reading of Michelman is that
demoralization costs have long played an implicit role in the development of
compensation law, but that the law has suffered from the lack of an explication
of this concept. Michelman's purpose and mine is to correct this condition by
providing the relevant organizing framework.
That demoralization costs can be interpreted as an extension to conventional
externality analysis must be regarded as a distinct gain. Although the economics
profession long questioned the operationality and empirical significance of the
externalities concept, I would judge that this has been resolved affirmatively in
both respects. Recent refinements and applications of the theory have contributed
to this result. Subsequent refinement and experience with the demoralization
cost concept should improve its power also.
Even without these, and confining attention to the papers reported in this
volume, the concept has relevance that extends beyond the World Trade Center
situation. It can also be brought to bear on the uncompensated urban highway
and urban renewal spillovers examined by Downs.
Administrative Decision Making and Pricing
131
Center construction is deliberate, and even if only at a late stage the
extent of its spillover effects became apparent, design changes might
still have been made—albeit that the city could not itself require these.
Thus criterion 1 is satisfied in large measure.
Criterion 2 is concerned with the administrative costs of securing
settlement. Is it realistic to expect in this instance that the identity
of and effect on disadvantaged parties can be determined at low cost?
With respect to the identity question, it must be conceded that genuine
differences
exist. Two engineering studies sponsored by the Port
Authority indicate that the cone of objectionable interference will be
3 to 5 degrees wide, while a third study places the cone at 30 degrees.
Which of these is more nearly correct will obviously affect the optimal
design. But it should nevertheless be possible to establish objectively—
e.g., with a test set brought into each neighborhood—the extent of
interference experienced ex post. Such an approach would be consistent with the objective observer standard proposed by Michelman.34
Also, it should be noted, precise satisfaction of every claim is not
necessary for compensation to be effective. A combination of an objective test of interference with an "average" compensation payment
would seem in this instance sufficient to drive administrative costs to
relatively low levels.35
With respect to criterion 3, the extent of harm experienced, there
is ample reason for regarding a spillover of from $2 million to $13
million as serious. As the Task Force puts it "there is no doubt that
the television set is a constant source of professional entertainment
and a constant window on the significant events that occur in the
nation and throughout the world. This is particularly true for those
who live in the deprived sections of northern Manhattan and the
Bronx, which will lie directly in the path of the cone of objectionable
interference caused by the construction of the twin towers of the World
Michelman's objective observer loss-estimation standard is that which we
would "impute to ordinarily cognizant and sensitive members of society" (op.
cit., pp. 1215—16). Otherwise estimation is
relationship.
apt to degenerate into a gaming
As the extended version of the model makes clear, compensation is better
regarded as a matter of precision than in either-or terms, As indicated in the
text, if the administrative expense of paying compensation is subject to a large
fixed cost, optimal compensation may be to pay nothing; but generally, where
significant secondary responses can be anticipated, compensation to victims in an
amount equal to the expected value of the loss would appear indicated. The
degree of precision attempted turns not on the mean compensation paid, but
on the deviation in the value of actual loss from estimated.
I
132
The Analysis of Public Output
Trade Center."36 Those who experience the interference will therefore
make the expenditures necessary to receive the directional UHF signals
or, lacking resources, will be denied an important source of satisfaction.37
Criterion 4 deals with the intrinsic merit of the project. There are
those who insist that the World Trade Center is even lacking in this
respect. I will assume for our purposes here, however, that it qualifies
as a project having positive net gains at design size X0—at least if
spillover costs are neglected.
The primary objectives specified by the Port Authority for the
design of the Trade Center are relevant in considering criterion 5.
These are: "that the Center serve as a symbol and focus for the
interest and involvement of the Port of New York in international
trade and commerce; that the Center meet the highest aesthetic
standards and be a pleasure to eye and spirit; and that the Center
provide ten million square feet of rentable exhibition and office
space."38 Families living in Harlem and the Bronx are not obviously
the principal beneficiaries of such an undertaking. Such indirect employment benefits as result would probably obtain for any building
providing ten million square feet of useful space. One must conclude
that evident reciprocities between burden and benefit are missing.
Finally consider criterion 6—whether substantial numbers of those
required to bear the spillover costs have what they regard as preexisting grievances against society, where these have a recognized
legitimacy. Families living in the deprived neighborhoods of Harlem
and the Bronx almost certainly fulfill the conditions of this criterion.
With the single exception of criterion 4, and possibly even here as
well, a consistent reading is obtained on each of the criteria: positive
demoralization costs, possibly nontrivial in magnitude, are to be
expected if compensation is not paid. Against this must be weighed
the administrative costs of paying compensation—which, in this
instance, are evidently low. The crude quantitative estimate that
36 Task Force Report, op. cit., p. 25.
Conceivably the cost of bearing the spillover will be shifted back on the
landlords, who will either install the necessary antenna or experience reduced
rentals. This, however, I find doubtful. Landlords must be anxious to forestall
relocation by their tenants for this to transpire. For one thing, rent controls in
New York City may cause landlords to welcome rather than resist decisions to
relocate. Furthermore, even in the absence of this issue, threats to relocate may
not be viable. The interference is a transitory phenomenon; relocation is an
expensive response to it.
38 Task Force Report, op. cit., p. 3.
Administrative Decision Making and Pricing
133
emerges, given the apparent order of magnitude difference that separates these two factors, is that compensation is clearly indicated—at
least at the design size of 1,350 feet.39 Whether compensation would
continue to be necessary were the building redesigned to take explicit
account of the interference costs is unclear. But compensation here
involves more than a simple transfer: allocative efficiency considerations reinforce equity arguments in its favor.
Thus, although secondary responses of the protective variety are
probably not to be expected in this instance (coaxial cable service
may not be available or sufficiently attractive, and this may substantially exhaust the range of "long-term," anticipatory, defensive
investment responses to TV signal interference of the type described),
the expressed social discontent (aggressive secondary response) is
less easy to dismiss. The evidently low administrative expenses that
would be incurred in paying compensation would appear to be well
below the expected value of the demoralization costs of this reactive
type.
3. Centralization versus Decentralization
The inability of the Task Force to require either redesign of the World
Trade Center or compensation for damages caused its chairman to
propose an alternative administrative decision-making procedure:
"there must be within our system of public administration a single
focus empowered to look at and to weigh, one against the other, all
the implications of a large project." Conceivably the prevailing political
and legal realities make this the only feasible solution. The relatively
high costs of moving to a centralized decision-making system of this
sort should, however, be appreciated.4°
It could be argued, of course, that the situation described above is
not typical. It might be claimed, for example, that ordinarily, whenAlthough Steiner agrees with my evaluation that compensation should be
paid under the criteria proposed, he indicates that "for some other list, compensation should not be paid." Agreed. Suppose, for example, each of the
criteria were replaced by its opposite. The judgment on compensation would
then be reversed. But unless Steiner proposes such a change, which I find doubtful, or is explicit on other criteria he has in mind, his observation lacks operational significance.
40
Some
of these are discussed in my "Hierarchical Control and Optimum
Firm Size," Journal of Political Economy, April 1967, Vol. 75, pp. 123—38. See
also Gerald Sirkin, The Visible Hand: The Fundamentals of Economic Planning,
New York, 1968, Chap. 4.
-J
The Analysis of Public Output
134
ever social goals and bureaucratic goals are in conflict, differences are
resolved in favor of social goals. But this requires stewardship behavior
of an unusual sort. I suggest that it is more realistic instead to recog-
nize that there exists a "bureaucratic cost" counterpart for the term
private cost in the usual social cost versus private cost distinction. To
proceed on the supposition that spillover costs for which no compensation is required will, nevertheless, be fully reflected in the decision
making of public or quasi-public agencies appears, at the very least,
to be unnecessarily hazardous.
The preferred economic solution to this (and similar) administrative decision-making problems is to supply the agency with the
relevant set of pricing signals. The stipulation that compensation
must be paid if demoralization costs can reasonably be expected to
exceed the administrative costs of paying compensation should tend
to induce both superior designs and a preferred mix of programs. The
above analysis is thus but another example of how, under the appropriate institutional rules, the use of prices can be made to help solve
complex economic issues in a decentralized way.4'
IV. Conclusions
Four principal conclusions emerge. First, with respect to the par-
ticular spillover condition examined above, allocative efficiency considerations would appear to support the payment of compensation for
the prospective television interference that will result from the World
Trade Center construction. This judgment is reinforced by appeals
to equity. The failure of compensation to be required in circumstances
of this sort strongly suggests that the social decision-making apparatus
is defective.
Second, evaluating the efficiency implications of spillovers of the
sort considered above requires that the concept of externalities be
expanded to include costs which take the form of secondary adaptive
responses (demoralization costs). These need to be weighed against
the administrative costs of paying compensation. The latter costs are
apt to be great, and hence compensation may be prohibitively expensive, if it is stipulated that, whenever compensation is made, damages
41
The
tax implications of this compensation argument would presumably
require that it be qualified. An examination of this question is beyond the scope
of the present paper.
Administrative Decision Making and Pricing
135
shall be ascertained exactly. However allowing as we have for "imper-
fect" compensation restores the likelihood that, absent large fixed
administrative expenses, allocative efficiency will support compensation whenever nontrivial secondary responses to spiflover conditions
are reasonably probable.
Third, although the emphasis throughout has been on allocative
efficiency, one should not suppose that an essential conflict with
equity exists. As Michelman has observed, the same general criteria
as are invoked under an allocative efficiency standard will ordinarily
be operative in reaching an equity judgment. This applies not merely
to the question of whether to compensate but also to the matter of
precision: one would expect that the conditions under which demoralization costs are especially sensitive to the exactitude of compensation would also be ones for which equity demands precision.
Finally, an issue that has been exposed but incompletely examined
in the above discussion warrants acknowledgement. This concerns
the long-run rule-making implications of the theory, which is the
principal interest of the law. To evaluate each case separately would
clearly have serious administrative cost implications and would subject decision makers to excessive uncertainty. The law is therefore
concerned with the design of policies to deal effectively with general
classes of damage exposure in the long-run. If, as appears to be the
case, allocative efficiency analysis has reached a higher stage of development than has the evaluation of equity, and if, in addition, the conflict between these two criteria in individual cases is rarely substantial,
demoralization cost analysis would appear to have special relevance
for the rule-making process. But note in this connection that if interdependencies between rules are significant, effective rule making may
require a pre-ordering or simultaneous determination of issues.
Development of this aspect of the argument, however, is beyond the
scope of this paper.
-J
COMMENT
by PETER 0. STEINER, University of Michigan
I am flattered by the fact that Williamson, apparently in response to
my discussion, has both shortened his paper and revised it in such a way
that my major criticisms of his original paper are no longer relevant.
I am embarrassed only by the fact that what remains of my discussion
paper is so short as to suggest either sloth or acquiescence on my part.
The five numbered paragraphs immediately below are what is left of
my original comment. Since Williamson chooses to discuss my comments (in his footnotes 33 and 39), I will respond briefly in the last
two paragraphs.
(1) I have contemplated the phenomenon known as Oliver Williamson for some time and offer the hypothesis that he is not an individual
at all, but rather a committee. This is consistent with his enormous
productivity over the last several years, most of which is superb, and
also, less flatteringly, with this paper. Here he brings, again with
enormous energy and diligence, a large body of theory to bear on the
interference caused to the reception of television signals by the construction of the World Trade Center. My complaint is that his appetite
has run somewhat ahead of his digestion—that the real data about
the policy questions does not support as much apparatus as he brings
to bear on them, nor (conversely) is the theory sufficiently original or
powerful to be a major contribution in its own right.
(2) The central fact reported by Mayor Lindsay's task force appears
to be that the Port of New York Authority is designing and building
two towers which will cause massive TV interference to several hun-
dred thousand homes for two years. It is clear that the Authority
made its decision without contemplating this adverse effect. Williamson suggests that it is possible (and perhaps likely) that had the
Authority been obliged either to avoid the damage or compensate
those damaged, it would have redesigned its facility.
(3) The example is interesting in several ways. It provides illustra-
tion of the well-known theorem that if a producer does not pay the
marginal cost of a resource (in this case, altitude), he is likely to
overuse it. Second, it reminds us that the mere fact that the producer
.4
r
Administrative Decision Making and Pricing
137
is a public agency provides no protection against a divergence between
"private" (i.e., producer's) and social costs. Third, it does so with
particular force since the damaged parties are mainly members of
relatively underprivileged groups in our society, groups whom we are
particularly reluctant to disadvantage further.
(4) All of this is both useful and suggestive. Williamson's interest
in the case is twofold. First, as a vehicle for insightful application of
Michelman's concept of demoralization cost—a particular form of
externality-induced response that may be socially costly. I found the
concept of substantial interest in and of itself, but not especially helpful in discussing the World Trade Center. The reason is evident in
the maximizing equation. In equation 1 the objective function is given
as V = G(X) — S(X) — mm (D(X), A(X)}. But the essence of
the criticism of the Port of New York Authority is not its neglect of
the relative sizes of demoralization costs, D(X), compared to the
administrative costs of compensation, A (X), but rather its neglect of
S(X), the spillover cost. Nothing in the case, as developed, measures
or identifies the demoralization costs or gives a clue as to whether this
is an operational concept.
(5) Williamson's second major interest is in suggesting rules for
when compensation should be paid. To the five criteria suggested by
Michelman, he adds a sixth. My objection here is not to this list of
normative judgments, but to Williamson's apparent failure to realize
that it is wholly self-contained. Given this, it follows that compensation ought to be paid in the World Trade Center situation. For some
other list, compensation should not be paid. These criteria are not
derived from the theory, nor do they enrich it. Williamson has neglected
to integrate them into his analysis.
My query in paragraph 4 is not whether demoralization cost is an
operational concept, but whether the analysis of the World Trade
Center case sheds any light on the operationality of the concept. I
believe Williamson has made a contribution in extending and refining
Michelman's rather loose concept. Granting this, does his review of
the World Trade Center case attempt to measure the size of the
demoralization cost, or merely illustrate anecdotally the concept? I
continue to believe it is the latter.
My query in paragraph 5 seems so clearly stated that I am astounded
that Williamson misses the point in his footnote 39. Suppose I were
to say that compensation should be paid whenever (a) those damaged
had legally enforceable claims, or (b) had incomes that were below
-j
I
138
The Analysis of Public Output
the national average. These are value judgments that are not without
support in our society, but they do not derive from the MichelmanWilliamson analysis. Thus they provide an answer to the compensation
question that is independent of that analysis. The same is true of
Williamson's six criteria.
—